add support for fonts inline in css as url data:

   The handler parses the first data value and has support for only
   one exact entry, but it can be adapted for more.

1 files changed, 72 insertions, 39 deletions

diff --git a/savewebfonts_lib.py b/savewebfonts_lib.py
index 53615c1..b0a6f30 100755
--- a/savewebfonts_lib.py
+++ b/savewebfonts_lib.py
@@ -17,7 +17,7 @@
 #    req-devuan: python3-bs4, python3-tinycss2
 #    rec-devuan: python3-fonttools, eot2ttf
 
-import requests, os, json, tempfile, subprocess
+import requests, os, json, tempfile, subprocess, base64
 from sys import stderr
 from bs4 import BeautifulSoup as bs # python3-beautifulsoup4
 from urllib.parse import urljoin, urlparse
@@ -138,61 +138,88 @@ def save_font(url,destdir,config):
    filename=""
    filename=os.path.basename(urlparse(url).path)
    ext = os.path.splitext(filename)[-1]
+   tf = None
    # Do not try to convert .svg
    if config.convert and not filename.endswith(".ttf") and ext not in [".svg"]:
       need_convert = True
       orig_filename = filename # in case we cannot load library later
       filename = ttfify_filename(filename)
+
+   if url.startswith("data:"):
+      if url.startswith("data:application/x-font-woff;charset=utf-8;base64,"):
+         need_convert = True
+         ext = ".woff"
+         tf = tempfile.NamedTemporaryFile()
+         contents = url[len("data:application/x-font-woff;charset=utf-8;base64,"):] # no worries about dryrun; we have already downloaded the font contents which are inline in the css file itself.
+         tf.write(base64.b64decode(contents))
+         filename = ttfify_filename(contents[:20])
+
    filepath = os.path.join(destdir, filename)
 
    if not os.path.exists(filepath):
       if url.startswith("data:"):
-         # not supported!
-         # WORKHERE: support saving to a tempfile this datastream, probably a base64encoded woff file. Then just convert.
-         eprint(f"Warning: Url {url[:config.MAX_STRING_PRINT_LENGTH]} is unsupported.")
-      else:
-         if not config.dryrun:
-         # Download content
-            response = config.session.get(url)
+         # Yes, some repetition here.
+         if url.startswith("data:application/x-font-woff;charset=utf-8;base64,"):
+            pass
+         else:
+            # not supported yet!
+            eprint(f"Warning: Url {url[:config.MAX_STRING_PRINT_LENGTH]} is unsupported.")
+            return -1
 
+      if not config.dryrun:
+         if tf:
+            with open(tf.name,'rb') as otf:
+               file_contents = otf.read()
+         else:
+            # Download content
+            response = config.session.get(url)
             if 'Content-Disposition' in response.headers:
                filename=response.headers['Content-Disposition']
                eprint(f"Using content-disposition value of {response.headers['Content-Disposition']}")
                if need_convert and not filename.endswith(".ttf"):
                   orig_filename = filename # in case we cannot load library later
                   filename = ttfify_filename(filename)
-               filepath = os.path.join(destdir, filename)
-
-         try:
-            if config.debuglevel >= 1:
-               sstring = "Saving" if not config.dryrun else "Save"
-               eprint(f"{sstring} {url} to file {filepath}")
-            if not config.dryrun:
-               if not need_convert:
-                  with open(filepath,'wb') as thisfile:
-                     thisfile.write(response.content)
+            file_contents = response.content
+
+      filepath = os.path.join(destdir, filename)
+      #try:
+      if True:
+         if config.debuglevel >= 1:
+            sstring = "Saving" if not config.dryrun else "Save"
+            eprint(f"{sstring} {url[:config.MAX_STRING_PRINT_LENGTH]} to file {filepath}")
+         if not config.dryrun:
+            if not need_convert:
+               with open(filepath,'wb') as thisfile:
+                  thisfile.write(file_contents)
+            else:
+               # need_convert is true, and not dryrun, so call function
+               if ext in [".woff",".woff2"]:
+                  try:
+                     from fontTools import ttLib
+                  except Exception as e:
+                     raise e
+                  convert_in = url
+                  if tf:
+                     convert_in = tf.name
+                  convert_woffwoff2_ttf(convert_in,filepath,config=config)
+               elif ext in [".eot"]:
+                  convert_eot_ttf(url,filepath,config=config)
                else:
-                  # need_convert is true, and not dryrun, so call function
-                  if ext in [".woff",".woff2"]:
-                     try:
-                        from fontTools import ttLib
-                     except Exception as e:
-                        raise e
-                     convert_woffwoff2_ttf(url,filepath,config=config)
-                  elif ext in [".eot"]:
-                     convert_eot_ttf(url,filepath,config=config)
-                  else:
-                     # no plan for conversion!
-                     eprint(f"Warning: no conversion plan for ext {ext} of {url}. Saving as-is.")
-                     with open(filepath,'wb') as thisfile:
-                        thisfile.write(response.content)
-            return 0
-         except Exception as E:
-            eprint(f"Error when downloading {url}, {E}")
-            return -1
+                  # no plan for conversion!
+                  eprint(f"Warning: no conversion plan for ext {ext} of {url[:config.MAX_STRING_PRINT_LENGTH]}. Saving as-is.")
+                  with open(filepath,'wb') as thisfile:
+                     thisfile.write(file_contents)
+         if tf: tf.close()
+         return 0
+      #except Exception as E:
+      #   eprint(f"Error when downloading {url}, {E}")
+      #   if tf: tf.close()
+      #   return -1
+      if tf: tf.close()
    else: # filepath does exist
       if config.debuglevel >= 2:
-         eprint(f"File {filepath} exists for {url}. Skipping.")
+         eprint(f"File {filepath} exists for {url[:config.MAX_STRING_PRINT_LENGTH]}. Skipping.")
+      if tf: tf.close()
       return 0
 
 def get_all_fonts_from_csslist(all_css, config):
@@ -294,7 +321,13 @@ def convert_woffwoff2_ttf(url, filename, config):
    Save the given url to filename, with filetype ttf
    """
    # This will only be called from save_font when dryrun=False, so the dryrun flag here is useful only if called from some other usage.
-   response = config.session.get(url)
+   if (url.startswith("http://") or url.startswith("https://") or url.startswith("ftp://")):
+      response = config.session.get(url)
+      file_contents = response.content
+   else:
+      # assume local file
+      with open(url,'rb') as o:
+         file_contents = o.read()
    try:
       from fontTools import ttLib
    except ModuleNotFoundError:
@@ -304,7 +337,7 @@ def convert_woffwoff2_ttf(url, filename, config):
       raise e
 
    with tempfile.TemporaryFile() as tf:
-      tf.write(response.content)
+      tf.write(file_contents)
       font = ttLib.TTFont(tf)
    if config.debuglevel >= 3:
       eprint(f"Converting {url[:config.MAX_STRING_PRINT_LENGTH]} from {font.flavor} to ttf as file {filename}")